- author = "C. J. Veenman",
- title = "Positional Genetic Programming",
- school = "Dept. of Mathematics and Computer Science",
- year = "1996",
- address = "Leiden University, The Netherlands",
- month = nov,
- keywords = "genetic algorithms, genetic programming",
- broken = "http://www-ict.its.tudelft.nl/~cor/thesis96.ps.gz",
-
URL = "
http://citeseer.ist.psu.edu/cache/papers/cs/2905/http:zSzzSzwww-ict.its.tudelft.nlzSz~corzSzthesis96.pdf/veenman96positional.pdf",
-
URL = "
http://citeseer.ist.psu.edu/veenman96positional.html",
- size = "66 pages",
-
abstract = "In this report a number of new reproduction operators
for genetic programming (GP) is introduced. These
operators have in common that they link the genetic
information to a position in the expression or program
tree. The GP engine, that implements the operators,
limits both the operator arity and tree depth. In this
way the expression trees become restricted structures.
The information linking together with the fixed
structures, transform the GP engine into a genetic
algorithm with even better performance than common
GP.
Without loss of effectivity, these adjustments to ordinary GP have the schema theory apply well to GP.",
-
notes = "
Positional Genetic Programming (genetic algorithms with encoded tree structures), 1996. This report describes a genetic programming engine implemented as a genetic algorithm, i.e. it uses fixed sized chromosomes and allows for ordinary crossover operators. Positional Genetic Programming turns out to perform better than the usual genetic programming scheme, that uses variable sized structures and random subtree exchange.
See also \cite{eiben:email:10-Nov-1997} thesis96.ps.gz doesnt work with ghostview 6-apr-98",
